Posts
Comments
Quick thoughts:
I'd say it looks a shade long, but I could well be wrong about the length of survey people will answer. Some suggestions for cutting it down a little:
- Questions 2-4 in section 1 seem somewhat redundant with one another to me (i.e. you could probably have just one or at most two of them).
- The list in question 1 (section 2) seems long to ask people to rate all of. Could you drop a few? (I'm thinking "self-help" is too broad, epistemics & uncertainty could maybe be merged, etc.).
You might also want to ask people to rate different parts of the course (moderation, content, structure, etc.) so you have an idea of what needs improving.
Overall, looks good! Thanks for running the project, strongly believe that evaluation is a really important part of any course/organisation/whatever.
"This is an internal document written for the LessWrong/Lightcone teams. I'm posting as "available by link only" post to share in a limited way, because I haven't reviewed the post for making sense to a broader audience, or thoroughly checked for sensitive things."
This post appears in search results and to people who have followed you on LW. I didn't read it, but you may want to take it down if this is unwanted enough.
ShareX does look like a more powerful (for some use-cases) version! I think the key benefits of Loom are it's extreme ease of use & its automatic upload of the video, which makes sharing feel very streamlined.
Unfortunately, I'm on macOS currently, so I can't test ShareX myself.
Really great post! The concept I have in my head looks broadly-applicable though slippery
The section below sounded a lot to me like "you form a model from a set of words, and then later on you Directly Observe the Territory™, and this shifts the mental model associated with the words in an important way".
Running on this model, I think a lot of the sequences was like this for me-- it wasn't until 1-2 years after reading them that I noticed concrete, major changes in my behaviour. Possibly this time was spent observing the part of the territory I call my brain.
But in fact, there really is a kind of deeper, fuller, contextualized understanding, a kind of getting-it-in-your-bones, that often doesn’t show up until later. Because when you first hear the wisdom, it doesn’t really matter to you. You’re usually not in the sort of situation where the wisdom applies, so it’s just this random fact floating around in your brain.
Often, it’ll be years later, and you’ll be in the middle of a big, stressful situation yourself, and that little snippet of wisdom will float back up into your thoughts, and you’ll go “ohhhhhhhh, so that’s what that means!”
You already knew what it meant in a sort of perfunctory, surface-level, explicit sense, but you didn't really get it, on a deep level, until there was some raw experiential data for it to hook up to.
Thanks for running this survey! I'm looking to move into AI alignment, and this represents a useful aggregator of recommendations from professionals and from other newcomers; I was already focussing on AGISF but it's useful to see that many of the resources advertised as 'introductory' on the alignment forum (e.g. the Embedded Agency sequence) are not rated as very useful.
I was also surprised that conversations with researchers ranked quite low as a recommendation to newcomers, but I guess it makes sense that most alignment researchers are not as good at 'interpreting' research as e.g. Rob Miles, Richard Ngo.
I think "speech of appropriate thought-like-ness" is, unfortunately, wildly contextual. I would have predicted that the precise lengthy take would go down well on LW and especially with ACX readers. This specific causal gears-level type of explanation is common and accepted here, but for audiences that aren't expecting it, it can be jarring and derail a discussion.
Similarly, many audiences are not curious about the subject! Appropriate is the operative word. Sometimes it will be appropriate to gloss over details either because the person is not likely to be interested (and will tune out lengthy sentences about causal models of how doctors behave), or because it's non-central to the discussion at hand.
For instance, if I was chatting to a friend with a medical (but non-rationalist) background about marijuana legalisation, the lengthy take is probably unwise; benzodiazepines are only peripherally relevant to the discussion, and the gears-level take easily leads us into one of several rabbit holes (Are they actually unlikely to cause withdrawal symptoms? What do you mean by unlikely? Does psychological addiction mean precisely that? Is that why those guidelines exist? why are you modelling doctors in this way at all, is that useful? should I be using gears-level models?).
Any of these questions can lead to a fruitful discussion (especially the last few!), but if you have specific reason to keep discussions on track I would save your gears-explanations for cruxes and similar.
This is good for some formats; I think in verbal communication I like to track this because the key variable I'm optimising on is listener attention/time; giving both loses a lot. I find it can be useful to save the gears-level stuff for the cruxes and try to keep the rest brief.
I strongly agree with the Johnswentworth's point! I think my most productive discussions have come from a gears-level/first-example style of communication.
What I'm arguing in this post is very much not that this communication style is bad. I'm arguing that many people will stop listening if you jump straight to this, and you should explicitly track this variable in your head when communicating.
Obviously 'know your audience and adjust complexity appropriately' is quite a trivial point, but to me thinking about it with a 'thought-like-ness' frame helps me to actually implement tis by asking "how much translating do I need to do for this audience?"
Maybe I should rewrite the post as "Gears in Conversation" or so.
I think it's good to experiment, but I actually found the experience of being on the site over the last week pretty unpleasant, and I've definitely spent much less time here. I initially went through some old ideas I had and tried posting one, but ended up just avoiding LessWrong until the end of the week.
I'm not totally sure right now why I felt this way. Something-like I'm very sensitive to feeling like my normal motivation system is being hijacked? I spent all of my time thinking about the best way to act differently given GHW, rather than just reading the content and enjoying it. This was pretty uncomfortable for me.
I'm sure this happens in many areas (maths, for one), but medical language is a pretty well-optimised system I know well. You might like to use it for inspiration:
Medicine: "72yo F BIBA with 3/7 hx SOB, CP. Chest clear, HS I+II+0. IMP: IECOPD"
English: 72 year old woman brought in by ambulance because she's been short of breath and had chest pain for the past 3 days. No noises were audible over her lungs with a stethoscope, both of her heart sounds were clearly audible with no added sounds. I think it's most likely this is being caused by an infection on top of a history of chronic obstructive pulmonary disease.
Guess who read more about what exactly people are pointing at when they say 'be agenty' and figured out that's what I'm trying to point at! That's right, it's me. Post cancelled, everybody go home.
Good post! Stop assuming things are/aren't good and go and look.
Is it worth making live-call-reviews a feature request for the LW feedback system? (Possibly limited to higher karma than the 100 required for text feedback, as I imagine this would have a smaller bottleneck with timezones etc.?). I imagine this would encourage a lot more people to say "I’ve got to do this now!".
If they’re interested in studying confusion, I ask them to tap their leg every time they notice they’re confused.
I tried this! It was enlightening. I didn't realise it, but I don't quite understand what my 'confusion' label is actually pointing at. I found myself confidently tapping my leg and then pausing, unsure of whether what was going on was truly confusion.
After a couple of days of this, what I think is going on is that I both did and didn't have separate labels for 'ignorance' and 'predictive error'. Some part of me was confidently tapping my leg whenever I didn't know something, and another part was saying "I don't feel confused, though. I just don't know, and I know that I don't know".
I already knew intellectually that the territory is stranger than I give it credit for, but I think this is one of the best examples of observing that first hand. It's more qualitatively different to listening to somebody say "Here's a label that confuses me" than I would ever have dreamt.
Basically any "Beginners set" online should set you right as a cheap way to try it out (though most pros say it's not worth your time. YMMV).
You will probably find it's hard with cheap shitty picks, but you'll (for fairly cheap) get a feel for which different shapes do what, and which you find most intuitive/useful. If you find it useful/fun, you could then either scour Ebay for cheap locks (with no guarantee of ease, but locks-without-keys is niche enough you can sometimes get good deals) or buy an Abus 45 or Masterlock. If I remember correctly, both of those should be 4 or 5-pin with no security pins, so about as easy as real locks get.
Once you feel you've graduated from your beginners set and want to splash some cash, you probably want to pick 1-2 pick shapes you got on well with and get really nice ones, or get a nice pick set. Here are some well-regarded vendors. You might also want to look into a practice lock, which seems like pretty good value for money.
Regarding learning, Reddit is pretty good for this one (as seems to often be the case with metis-skills). In particular, the Belts stuff seems like a pretty good curriculum. I don't expect sitting reading about lockpicking will be very useful compared to, well, picking locks (this is the whole point of this post!).
If you want to feel inspired (and never trust a lock again), LockPickingLawyer is a personal favourite with some educational stuff. He also has an excellent sense of comedic timing.
Let me know what you found useful & how you get on!!
This is exactly what I'm saying. Using machines in ways they're not made for is especially risky when the machine controls access to your house.
I'm not sure if this tag should be about the general concept of past and future selves, or about coordination problems with past and future selves & TDT. Either seems valuable to me, but it seems like the latter was intended at creation, so I've continued in that vein.
I am also unsure of exactly what it is, but I used to fairly consistently induce a similar feeling in myself with 'mindful walks', also inspired by Original Seeing. For me, it was closely bound up with getting curious about things I'm used to looking at without seeing-- what are those marks on pavements? A lichen? Are they raised above the pavement? What do the different colours and shapes look like? Why are they round-ish and spaced out, rather than covering the whole surface, or some other shape?
This might not be 'true' curiosity-- I never looked up other people's maps for an explanation of the marks on pavements, for instance-- but it does fairly often give me a feeling of 'realness'. I was struck by how similar your not-a-SIM-key experience was.
Your example of a friend saying 'let me be real with you' reminded me of the concept of the 'press secretary'. Asking the right questions, and having a certain emotional quality, seems to trip up my press secretary for long enough that I can query the things behind her a little.
EDIT 19/03/22: Maybe in future I should finish sequences before I leave comments on them
"The thing about those distinctions is that they are a) useful, and b) curiosity-stoppers. They tell us "don't worry, you already know this" so you can get back to building a tower of interconnected concepts. Which is a good thing, most of the time, but it is a bad thing some of the time"
I liked this footnote, but I'm not sure why. I'm going to say some things to try to think about it more clearly.
What this footnote seems to me to be about (in part) is something like:
- Stop attaching string to your insane-person cork board
- Notice that the things you are connecting with string are sketches, not photos
- Remain stopped for long enough to fill in your sketches a bit, erase some bits, and add a little colour
On this model, I am truly appalling at [1], and therefore rarely get the opportunity to practice [2] and [3]. I actually quite enjoy the feeling of remaining stopped on its own, but I think adding string feels to me too much like 'learning things' for me to look past it very often.
This model is (of course) wrong, but it feels closer to me than other words I have to point to it.
I also noticed that my brain likes first-things to cause (be required by?) second-things, so my initial model of the main text was something like familiarity → facts → identification → models → mastery. This could be intended, and does reflectively seem fairly sensible, but I can imagine having practical mastery over something without a complex (or even correct) model of it. Exercising seems like a good example where I think sufficient experience could create practical mastery without a strong model or many facts.
However, I was surprised when you picked out driving as an example, as I wouldn't have said I have a strong model of how a car works. This probably means I've misunderstood what you mean by 'models' and 'facts'.
I think what's going on is that I'm getting distracted by the context I usually hear the words 'model' and 'fact' in, next to words like 'science', 'engineering' and 'textbook'. This is getting in the way of me thinking about things like 'if I exercise when I haven't in a long time, my arms and legs will feel sore afterwards' as facts.
This elicited in me a very specific kind of joy I first experienced reading ZAMM, and for which I have made all too little time ever since. I have nothing substantial to say beyond that I find your prose delightful in the same way I find delight in Original Seeing. Thank you!
with hindsight, I can say with some confidence that this was roughly the right advice. Whilst I did need to read most of the university content (and fast), 'slow' is definitely the right way to engage with the most content on LessWrong. Thanks!
Disclaimer: I know essentially nothing about US legislation, scientific ethical frameworks, etc. as I am not American. I just read the paper and have some background in genetics.
tl;dr: No, this is classic gain-of-function research as far as I can tell
From the paper, I can see two vaguely plausible arguments for why this isn't gain-of-function research:
- SARS-CoV is already (obviously) already capable of infecting human cells. Using SARS-CoV as a vector to test other spike proteins' ability to infect humans doesn't increase the number of hosts
- Prior to the chimeric testing referred to, it seem like the authors did not expect the altered spike protein used (SHC014) was likely to successfully infect human cells; host-range regions were different from SARS-CoV, and it was unable to enter human cells when in a pseudovirus (which, as I understand it, are not fully replication-competent, making an outbreak unlikely)
However, neither of these arguments really holds water in my opinion. The first seems the strongest-- my main concern is that introducing a different spike protein could plausibly increase transmissibility or pathogenicity, but I don't know enough about that topic specifically to confidently evaluate that claim. If anybody does know I'd be interested to hear (for instance, do any SARS-CoV-2 variants have spike mutations?).
As to the second point; if you didn't think it was plausible that SHC014-MA15 (the chimeric virus) would be capable of infecting human cells... why did you do the test in the first place?
I think this should probably be merged with cognitive reduction, which is more general and (I think?) encompasses this one
From the old LessWrong Wiki Discussion Page:
Talk:Jargon
Phyg and Phygish
"Phyg" and "phygish" are used a lot. I'm looking for recommendations on how to define them without putting this page in the wrong Google index. --R claypool 15:03, 10 August 2012 (UTC)
What counts as jargon?
I've recently had an addition or two of mine removed form the jargon file that I disagree with. So let me explain why I've been adding them.
I'm happy to take the definition of jargon to be "the language, especially the vocabulary, peculiar to a particular trade, profession, or group"
Now, I'm a reasonably well-read lay-person, but every so often, when I'm reading a discussion in comments, I'll come upon a word that I have to go look up on wikipedia to understand it before I can figure out what the commenters are talking about.
I consider most examples of this happening to mean that they're using a word that is jargon. In most cases - the words I've not understood were philosophical jargon... ie you have to have studied at least a solid base of philosophy to understand what they mean without reaching for the dictionary.
I'd consider words such as utilitarianism, consequentialism and deontology to be good examples of such philosophical jargon. I might guess at what I think they might mean - but to be sure - a definition (and link to a better explanation) is a good idea to have on hand... and therefore I added them to the jargon file.
The reason being that: if a complete newbie (such as myself) doesn't understand them... then so will other newbies - and we are excluded unnecessarily from the conversation.
My argument is in favour of allowing these words in the jargon file for this reason.
Content of this article
Should this article be a list of Jargon with short descriptions or just repeat the contents of Category:Jargon? MrHen 16:42, 23 April 2009 (UTC)
Random idea: Make this article a list of Jargon with short descriptions, and transclude the contents of this article onto the category page. The reason for duplicating the content on the category page is because when browsing through the category trees, users will often end up on the category page, rather than the article page. --PeerInfinity 17:40, 23 April 2009 (UTC)
Unclear and hard to use
Unless there's overwhelming objection, I'm going to merge the acronyms list to this article and reformat it more like a list of short definitions for the newbie, like most jargon lists I've seen. (Certainly it shouldn't be spread over two pages as it is now, with this page not actually providing any explanations at all.) This should be a single-point info list for n00bs - David Gerard 09:38, 17 April 2011 (UTC)
No objections, so I've done this. Do of course feel free to fix any of my quick definitions you don't like - David Gerard 18:49, 21 April 2011 (UTC)
Hmm, ok. I think both courage and evolution I found more difficult because they're less well-defined clusters in postspace (compared to self-deception and superstimuli). I'm glad you found the feedback helpful.
I've also edited gears-level and spaced repetition. I think they're probably C and B class respectively, but I'm still very unconfident about that. Gears-level in particular I'm not sure if it might not just be better to point to Gears in Understanding, as it's pretty well-written and is pointing to an odd (& specific) concept.
Edited the courage tag, think it's C-class (Not sure if it needs integrating somehow with the groupthink and/or heroic responsibility tags? certainly some things in each of these don't fit under the others but there is a fair amount of overlap at present)
Edited self-deception & superstimuli, think they're now C-class (self-deception in particular, I'd like somebody who's actually read Elephant in the Brain to have a look over it, because it seems relevant but I'm not overly familiar)
Edited evolution and think it's now B-class
I'd imagine publication bias priors are helpful, especially with increasing specificity of research area, and especially where you can think of any remote possibility for interference.
Just as an example I'm familiar with (note this is probably a somewhat more extreme example than for most research areas due to the state of pharmacological research): If you see 37 RCTs in favour of a given drug, and 3 that find no significant impact (i.e. 93% in favour), it is not unfounded to assume that the trials actually performed are roughly equal in favour and against, and that there may be a missing 34-odd studies.
A 2009 analysis found that this was almost exactly the case (the studies registered were 36:38 in favour of the drug; one positive RCT went missing before publication. Along with twenty-two non-significant studies that were missing altogether, and a further 11 which were so poorly analysed as to appear significant.
(Bad Pharma, by Ben Goldacre, is a pretty sound resource for this topic in general)
I've updated the Heuristics and Biases tag again btw. I don't think it's A-grade based on "I'd like to see more work done on it", but I think it's about as good as I personally am going to be able to get it. I'd really like somebody (yes you, fellow user reading this) to have a read through and make any adjustments that make sense and/or make it more comprehensive.
re: fallacies, I thought about it, and I think they're actually used pretty similarly, at least here on LW. Planning fallacy could easily be described as a bias generated by an 'imagine your ideal plan going correctly (and maybe add, say, 10%)' heuristic. At the very least, there's plenty overlap. Really what I envisioned for that section was making the point that a heuristic can be good (or just ok), because that was something that I didn't realise for a long time.
OK. So you see the grading as being more of a "neglected-o-meter" in the sense that it describes the gap between how a tag currently is and how it would be in an ideal world? (i.e. a more important tag would have a higher bar for being A-grade than a less important one?)
I think that makes more sense than an absolute-quality stamp, but I think the tag grading post as is currently written should make that clear (if it is the case)-- currently it implies almost the opposite, at least as I read it. For instance phrases like "It covers a valuable topic" in A-grade, and "tagged posts may not be especially good." in C-grade. To me these read as "quality/importance of topic and of posts are as important for grading as description".
I think actually the way you're describing tags now is more useful (for e.g. directing peoples attention for improving tags), but I'm not sure if it came across that way (to me) in the initial post. I would be interested to hear how other people read it.
This all seems like really helpful advice, so thanks! Multiple-pass reading is something I've made previous attempts at but need to find a way to properly remember to implement, especially for longer things (like, say, books).
I generally timebox specialist reading that has a near-term goal-- reading for university or for a specific paper. The big problem for me personally is that, as a jobless university student, there is definitely a temptation (worsened by lockdown and summer holidays) to let more generic reading expand until it fills my the spare time in my day with little structure. I think your comment has really helped me highlight that as an issue, so thanks.
Well around half of them are sources I'm currently using to write a paper, and some of the rest I'm reading in preparation for next year of university. But I think I probably could benefit from a little of what you outlined.
I've edited the Heuristics and Biases tag. I think it's probably A-grade (I'm still getting a handle on exactly what an A-grade tag should feel like though, honestly).
That said, I'd like it if somebody could check the specifics of the three definitions, because I'm actually not completely sure, and check that it scans ok.
For sure! I figured the team wouldn't have missed this, just wanted to give my two cents. For what it's worth I think the tagging system is actually really nicely implemented already; I feel like a kid in a candy shop with all these posts that were just thoroughly inaccessible to me until now.
I've edited the Effective Altruism tag pretty heavily, and I now believe it qualifies as A grade.
I've also edited the Epistemic Modesty tag, and think it's now C or B grade.
I'd also like it if the X-risk and S-risk tags are consistent with one another-- I propose that "S-risks (Risk of astronomical suffering)" and "X-risks (Existential risk)" is the best format.
Are there any plans to implement tagging of whole sequences? I understand that tagging the first post in a sequence has a similar effect, but it might be more productive in some instances to have, for instance Slack and The Sabbath as the top link under the slack tag, rather than the individual posts from this sequence appearing in an order based on relevance.
Obviously that then creates issues about whether you want posts that appear in sequences to also appear individually or not, and whether you want all sequences to be taggable or not, and so on. I'm not sure if these issues outweigh the benefits; even just on an admin-only basis, it seems like a helpful feature if we expect a significant user base who don't read the tag descriptions (the other place it might otherwise make sense to put sequence links).
(Other examples where this seems to me it might be more useful than the current method: Philosophy of language & A Human's Guide To Words, Group Rationality & The Craft And The Community. I imagine there are more)
This is true, and a mistake on my part (they don't bother with IFR in medical school, likely because it's not as relevant for day-to-day medicine as CFR). I'll update the post to try to explain the difference. Thanks a lot.
To your first point: my intuition is that ACE2 is far too small for the genome to pass through itself. ACE2 is an enzyme that's bound to the membrane-- it actually just cleaves angiotensin 1 to angiotensin 2 (hence 'angiotensin cleavage enzyme, ACE2). It does pass through the membrane, but it's not really a 'channel'-- it is simply localised to the cell membrane, and acts on substances extracellularly.
Enveloped viruses can enter cells in many ways (principles of virology chapter 5 is really excellent for this, if you're interested). It seems that SARS-CoV (the original outbreak) enters cells primarily how it is implied above-- simple, direct membrane fusion mediated by the ACE2 receptor. There is some speculation that it may under some circumstances be endocytosed (taken into the cell in a separate sphere of membrane) and then break free of the endosome (the bubble) in a pH-dependent way. Obviously this is further complicated by the fact that this is SARS-CoV-2 that we're really interested in, so I thought it would be best to leave it blank. You're right in thinking this process is similar to SER budding, though.
To your second point: I wasn't actually sure! I've done some research, but honestly I'm still not as confident about this as about the rest. As far as I can tell, for most viruses nucleocapsid shedding is either mediated by substances or organelles inside the cytoplasm-- ribosomes in particular, apparently, bind to the capsids of some viruses and destabilise them -- or is part of the process of receptor binding. Some viruses, for instance, seem to be able to leave their nucleocapsid behind with their envelope so it coats one side of the cell membrane.
Sorry I can't give a better answer, hope it helps!
What sources are governments using for decision-making?
The biggest impacts seem to me to be via influencing government. The UK government, for instance, is still very reticent to enforce widespread testing or mandatory quarantine. Their 'quarantine guidance' for households with symptoms looks like this, which seems patently foolish for a number of reasons.
Influencing governments' decision making is high-impact and potentially tractable via getting modelling and trial data to them. The UK Government publish their 'scientific basis for decision making' but it appears to be weeks out of date and unreferenced.
With that in mind, how do we get better decision-making information into government? What theory of change can we find for influencing policy makers? I believe this should be primarily targeted towards larger organisations and researchers who can have more direct influence, but may be useful for individuals as well.
Primarily or exclusively due to the thread:
- Bought a pulse oximeter
- Copper tape on phone, laptop hand-rests, and doorknobs and light switches (due to living in shared student accommodation) -- including recommending others do the same; two or three people I know have taken me up on this advice.
- Bought, and take daily, vitamin D pills
- More careful with packages (treat external packaging as if it is infective)
- Gathered enough food/medicine to be able to quarantine and look after self if required
Due to 'seeing the smoke' due mainly to LW posts (would likely have done otherwise, but much later)
- Came home to my parents to lower risk from being in a city
- Properly educated myself on the virus and the outbreak
- Self-quarantined, including minimising contact from my parents until 15 days asymptomatic living here
I'd like to suggest not using ibuprofen, or any other anti-inflammatory (NSAID or steroid: also includes aspirin, cortisone, etc. with the presumed exception of pre-prescribed steroids for e.g. asthma).
This is on the basis of this article from the BMJ. In summary, there are a handful of COVID-19 specific cases of young fit people becoming severely ill following ibuprofen use, combined with small studies on SARS-CoV and other illnesses.
Paracetamol should probably be used instead of ibuprofen/NSAIDs.
This seems like a cheap switch as they're likely roughly equivalent in symptom control excepting this effect. Please let me know if you see or read any reason this may be dangerous/untrue. (Is this an appropriate place to post this? Or does it belong in the generalised advice thread?)
This feels to me like part of the puzzle, as you say.
I think the other part is some quality of mind-like-ness (or optimising-agent-ness, if you prefer). People rarely attribute free will to leaves in the wind, despite their inability to accurately model their movements. On the other hand, many people do regularly attribute something suspiciously free-will-like to evolution(s).
I don't have a good idea how either of these two concepts should be represented, or attached to one another, though.
Does this mean, then, that it is not merely difficult, but mathematically impossible for any matter to ever reach 0 Kelvin? This would seem to violate Liouville's Theorem as stated here.
On one notable occasion I had a similar discussion about sound with somebody and it turned out that she didn't simply have a different definition to me-- she was, (somewhat curiously) a solipsist, and genuinely believed that there wasn't anything if there wasn't somebody there to hear it-- no experience, no soundwaves, no anything.